PJ 15KW Quirks

[wonka]

I have a 2020 model 15KW 48V inverter. After using it for a few months, I'm curious if the peculiar behavior I've noted is exclusive to my unit or if there are others with the same problem(s).  I have it connected to the breaker panel feeding the whole house.

Somewhere around 350W L1 output/load, the output causes the light bulbs in my house to flicker @ 60hz with varying intensity.  If you look away from the bulbs, the rooms seems as if it were lit by flickering candlelight. It doesn't seem to matter what the load is, I get the same results with many different loads regardless of whether they are resistive or reactive. 

Also, the idle current fluctuates from approx 35-80 watts at a rate of about 1hz.  With the fans off, you can hear the buzz/hum in the transformer changing with the same rate, mimicking the fluctuation of the idle current.

Has anyone else noticed similar behavior? Have you been able to correct it?

[Sid Genetry Solar]

Have I had this problem?  Absolutely.

This problem is entirely caused by the voltage regulation code in the CPU, which has no hysteresis.  It is either turning up the throttle, or turning down the throttle--which obviously will cause an oscillation.  Funny part is, the oscillation is on a 60Hz basis (+1 / -1 on the high wave, low wave), causing one of the two L_MOS to run hotter than the other.

Adjusting the voltage regulation knob seems to "adjust" this issue--but as it's caused by the firmware, there isn't much of anything that can be done from a hardware perspective to fix it.  And of course, the loading on the inverter "adjusts" it further.  The best I could do with hardware (small signal capacitors, etc.) was get it from a 15 or 30Hz big oscillation all the time, to a 60Hz small oscillation most of the time.

Genetry Solar does have (in-house) revised PJ firmware that corrects this problem.  (And both L_MOS stay about the same temperature.)  If you have a v4-v9 PJ control board (with the removable CPU), you could buy a WiFi board upgrade kit from Sean (which comes with a replacement reprogrammed CPU board)--this will allow you to tweak, tinker and adjust a plethora of settings in the inverter.  As well as fix the flickering issue.  And give PWM fan control based on temperature.  (FWIW: the WiFi board allows OTA updates, so I can provide an update to tweak/add features as necessary.)

[wonka]

Thanks for the reply!

I would love to install a WiFi board but i have v10.03 control board.  Is there a solution that will work in my unit?  I'm willing to experiment and guinea-pig if need be. I do have 20years electronics background (mainly RF and mobile 12vdc electronics) and a full bench w/ scope, sig gen, freq counter, spectrum analyzer, power supply, etc.   Before i found genetry I had planned to reverse engineer this inverter but I hate to re-invent the wheel as it looks like you guys have 'er pretty well figured out.

 

[Sid Genetry Solar]

16 minutes ago, wonka said:

I would love to install a WiFi board but i have v10.03 control board.

Sigh.  Unfortunately not an easy one.  To go the Genetry route with a 10.3 (unified) control board would require Sean shipping a replacement pre-programmed 10.3 control board out with a core charge (refundable upon return of the 10.3 board you have).  Also your inverter chassis probably doesn't have a spot to mount the WiFi LCD board...count yourself really lucky if you got one of the units that has a plate covering up the LCD cutouts.  That was our suggestion; I have no idea how many were produced that way (though I have personally seen 1).

Once I finally get these Genetry inverters fully finished, I plan to go back to PJ and update their design (save them some money in the process!)  The updated CPU code I'll provide them would support the WiFi board, so it would be a fairly simple addon at that point.

Also hoping to provide an upgraded drop-in LF Driver that will actually drive those FETs pretty cleanly, and solve a lot of the FET heat (even though the transformer is the weakest link).  Only so much I can do from day to day though 😉

I really enjoy how many customers digging into these units have significant electronics background and test equipment...that's always a bonus.

 

[Sid Genetry Solar]

I guess I should summarize: yes, it's possible to run the GS upgrade on a 10.3 PJ inverter--just up to you how much work you're willing to go through to make the upgrade.

[Waterman]

2 hours ago, Sid Genetry Solar said:

I guess I should summarize: yes, it's possible to run the GS upgrade on a 10.3 PJ inverter--just up to you how much work you're willing to go through to make the upgrade.

Couldn't you use one of the V 8 board kits and then go with the WiFi board. It would still require cutting the hole for the LCD unless he has the plate on the side.

[Sid Genetry Solar]

4 hours ago, Waterman said:

Couldn't you use one of the V 8 board kits and then go with the WiFi board.

Could be done--though if he doesn't have a v8.0 board laying around, then he might as well go with a v10.3 replacement board and the core charge.

[wonka]

Thanks, I will check with Sean about the 10.3 upgraded board.  I do have the cut-outs to install the lcd but I am not opposed to mounting it elsewhere as I have already removed the plate in order to pass the wires for my transformer through the holes. I suppose that I should mention, knowing the transformers are the weakest link here, I have removed mine and installed it in a bath of 4 gallons of mineral oil for cooling.  pics available if anyone is interested...

So far this has worked extremely well for me. So well, in fact, the fans have yet to turn on via thermal switch.  I have noticed the resistors on the mosboards run pretty hot (about 50* over ambient temp) though without some air movement, so I have the 2 of the fans on that end running at reduced speed to keep everything cool.  the other fans are connected to energize w/ the thermal switch.  

I've seen it mentioned before that disconnecting one of the 2 mainboards can be beneficial. Is there a significant reduction in losses or increase in stability (is it worth the effort)?  my MOS boards only have 4 Fets installed; if I were to disconnect a mainboard would you suggest adding the additional Fets and resistors? 

 

[Sid Genetry Solar]

26 minutes ago, wonka said:

Thanks, I will check with Sean about the 10.3 upgraded board.  I do have the cut-outs to install the lcd but I am not opposed to mounting it elsewhere as I have already removed the plate in order to pass the wires for my transformer through the holes.

So the WiFi board would only use one of those cutouts.  It is worth noting that the "output board" would need to have the parts (a transformer, connectors, and a few resistors) installed for voltage/current feedback to work.  It's hit or miss whether PJ included them on the board.

The WiFi board is intended to be mounted to the chassis, as there's over 50 wires that can connect to a fully loaded WiFi board.  Battery power...up to 6 fans (PWM) or 3 fans (2-wire)...up to 6 thermistors...plus the feedback and CPU communication cables.

 

26 minutes ago, wonka said:

I have noticed the resistors on the mosboards run pretty hot (about 50* over ambient temp) though without some air movement

I would presume that this is the snubber resistor (R3 on the short MOS boards, or R14 on the long MOS boards), not all the gate resistors?  (LF Driver is far too weak to get the gate resistors hot, haha!)

 

26 minutes ago, wonka said:

I've seen it mentioned before that disconnecting one of the 2 mainboards can be beneficial. Is there a significant reduction in losses or increase in stability (is it worth the effort)?  my MOS boards only have 4 Fets installed; if I were to disconnect a mainboard would you suggest adding the additional Fets and resistors? 

Disconnecting the 2nd mainboard is most beneficial on inverters using the older long mainboards (over 12" long) with 6x HY3810 FETs per MOS board.  The PJ LF Driver is far too weak to drive a single set of 6 FETs (easy to see with your 'scope--those gate drive signals are very rounded)...which is why adding a 2nd mainboard of those heavy FETs often tended to cause the inverter to run hot at no load.  "Memo, PJ: the sine wave should be on the output of the inverter, not on the FET gates" 🤣

With your setup having (most likely) 2 short mainboards (~6" long) and only 4 of the newer NCEP039N10 FETs per board, you're probably alright with the double mainboard setup.

 

26 minutes ago, wonka said:

the other fans are connected to energize w/ the thermal switch.

Gotta love that thermal switch.  Nothing like putting a 0.5A 30vDC rated switch on 3 amps of 55v fans...what could possibly go wrong??

[Sean Genetry Solar]

Upload pictures of the unit and we can tell you what will be needed for an upgrade. Obviously a full control board wifi board upgrade would work out of the box and would eliminate alot of the hardware found inside the inverter. The downside is you need to make 2 pin connectors for the fans to plug in as well as a the positive and negative supplies to the wifi board but it can be done with little fuss.

[wonka]

Here's a pic of the output board, doesn't appear to be completely populated, though I'm unsure exactly what is required for the GS board.  I can add whatever needs to be added, and I don't mind making connectors for fans and what not.  would really like to upgrade completely to the GS board and I'm quite comfortable working on this.

[Sid Genetry Solar]

So a complete GS board upgrade (WiFi board + Control board) would completely remove the need for an output board.  We have very limited stock of control boards right now...but we do have a few.  You would need to get in contact with Sean for an upgrade kit of sorts...and I'm going to have get working on writing an upgrade manual sooner than I thought 😉

I've personally done a full GS upgrade on a 48v 15kw PJ inverter that had the twin short mainboards (and a 10.3 PJ control board)...but it sure isn't a job for the faint of heart!  Parts were everywhere, and it looked like a total wreck there for awhile 😉.

Here's a photo of it after the upgrade:

FWIW I was quite pleased find that the PJ tranny in that inverter had a 32v primary--most PJ "48v" trannies seem to have a 36v primary, which gives you flat-topped AC waves with minimal load.

Your fully stocked bench sounds like it might come in somewhat handy...

[wonka]

I've had the whole thing apart and back together a couple times already, the only part i thought was a pain was trying to get the plastic standoffs under the main board to stay in place long enough to drop a screw in. My unit had a screw from the terminal strip (for the fans) rattling around when it arrived so I tore it apart for close inspection.  While I had it apart I put some heatsink compound on all of the FET's and touched up some questionable solder joints.

My transformer is an asl6.5 with 36v primary unfortunately.

 

[Sid Genetry Solar]

Now that I think about it, we did have to extend the transformer secondary wires to reach the GS control board...bit of a pain, especially as the trannies are wound with aluminum wire--which can't be soldered to without special equipment (that I don't have).  But if you relish a bit of an electrical challenge...well...you can contact Sean and see about a full GS upgrade kit.

[wonka]

I should be a step ahead of the game in regards to the transformer wires as I have mine connected externally in an oil bath.  Always looking for a challenge!

[Sid Genetry Solar]

20 minutes ago, wonka said:

Always looking for a challenge!

Challenge granted 🤣

I see the displays you've added to the front of the inverter in an attempt to get some sort of useful information out of it.  Methinks you'll drool over all the settings and data available on the Genetry Solar WiFi board...

[wonka]

17 minutes ago, Sid Genetry Solar said:

Challenge granted 🤣

I see the displays you've added to the front of the inverter in an attempt to get some sort of useful information out of it.  Methinks you'll drool over all the settings and data available on the Genetry Solar WiFi board...

Absolutely... already drooling at the thought of it!

[Sid Genetry Solar]

Not like you can use one of the 3 PWM fan channels together with an external 15K thermistor to one of the 6 thermistor channels to have the WiFi board automatically throttle up a circulation pump if the oil starts to get a little warm 😉

 

[wonka]

yep, thats the kind of thing that gets me excited.  I already have the pump but so far haven't needed it.  Not sure what summer temps will bring though; my installation is in an un-condtioned outbuilding.  Last fall, after I got it all put together, a 6kw load for an hour only brought the oil temp up to about 130*.  I was able to manage that pretty well with just natural convection currents and a fan, but the ambient temp was only 70*.  Our summers can be quite hot, sometimes 100+ for a week or longer, so the pump may end up being necessary.

One question I have regarding the GS board; how does it bode in the RFI/EMI department? I've come to accept that I will probably just have to live with the increased noise levels on my amateur radio gear as my inverter and the MPPT's (MSB) are pretty noisy under load  I am, however, curious if the board will be affected when I transmit on various bands, sometimes at power levels >500watts.  I operate HF, VHF and UHF within proximity of the inverter gear.  I've noticed a few times that the voltage regulation will take some wild swings under certain conditions when I transmit but I haven't been able to reproduce that very consistently so there is still some detective work to be done on my end. Various RF bypass caps in different locations have seemed to reduce the symptoms to moot at this time so I haven't dug much further.

[Sid Genetry Solar]

38 minutes ago, wonka said:

One question I have regarding the GS board; how does it bode in the RFI/EMI department?

You might be a very valuable test bench in that regard 😉.  Like any designer, I want to believe it's solid, but until tested in every situation, it's hard to say.  There's significant EMI inside the inverter chassis...and it's pretty stable, so I don't envision there being any unwarranted issues.

38 minutes ago, wonka said:

I've noticed a few times that the voltage regulation will take some wild swings under certain conditions when I transmit but I haven't been able to reproduce that very consistently so there is still some detective work to be done on my end

Have to say that I'm not surprised at all: There is no true AC-DC isolation in a PJ inverter.  If you look at the PJ LF Control board, you'll probably notice the 4x8 block of 32 resistors on the PJ control board...each one is 499K ohms.  These form 2 strings of resistors in series.  One side connects to one phase of the AC output (i.e. 120v).  The other side directly connects to DC ground and an op-amp circuit.

This means that the AC output signal has roughly 16 megohms impedance to the DC control circuitry...and as someone working in RF, you know how much of an antenna that could be.

Adding insult to injury, after this signal is "rectified" by an op-amp (diodes in the feedback path, etc.), the output is then run over another foot of wires to the front panel adjustment knob, before being fed pretty much directly into the CPU.  This is another huge path for noise to wreck havoc with the AC output feedback--not to mention giving a great "input" for the inverter to get into wild oscillations with itself (which is part of the cause of your opening comment).

Sometimes it's a wonder that it works at all.

 

The GS board setup is considerably different: at least for the current revisions of boards, I am using optoisolators to truly isolate the AC side of the inverter from the DC battery side.  Impedance is considerably lower: less than 200K into the optoisolators...and less than 4.7K impedance directly into the CPU...with no ridiculous "adjustment" cable going another foot to the front panel.  All voltage regulation/calibration adjustments, etc. are 100% handled via firmware...which has a surprisingly high resistance to EMI 🤪.

So with that in mind...I think the GS control board should be considerably better.

[wonka]

16 hours ago, Sid Genetry Solar said:

You might be a very valuable test bench in that regard 😉. 

Happy to oblige.

17 hours ago, Sid Genetry Solar said:

Sometimes it's a wonder that it works at all.

I see what you mean, that explains a lot.  Glad to hear about the opto's on the GS board.  It should be FAR superior.

I'm sold. It sure sounds like the GS board solves all of my problems and gives me the flexibility to tinker the way I'd like to.  I suppose the next step is to place the order.  What's the best way to go about ordering- Call? or via the genetry website contact form?  Do I need to request anything specific for my unit? 

[Sid Genetry Solar]

33 minutes ago, wonka said:

Glad to hear about the opto's on the GS board.  It should be FAR superior.

Checked the schematic...impedance to the CPU on the GS board is actually <1k.  The only pitfall with the optos is their (significant) temperature coefficient.  Fortunately, with some external thermistors, we can reduce the temperature error from +15v (i.e. 220 -> 235v) to +4v...but still, future GS board revisions will be switching to a small feedback transformer to solve this issue.  (Impedance: similar.)

37 minutes ago, wonka said:

It sure sounds like the GS board solves all of my problems and gives me the flexibility to tinker the way I'd like to. 

Well, if you need any further "tinkers", you're talking to the programmer 😉.  There's not much that can't be tweaked--as a matter of fact, Sean & I have been tinkering with the H-bridge dead time settings lately.  I can easily provide a "tech" update to add/tweak a function as needed.

 

38 minutes ago, wonka said:

I suppose the next step is to place the order.  What's the best way to go about ordering- Call? or via the genetry website contact form?  Do I need to request anything specific for my unit? 

TBH I'm not super sure...at the end of the day, you're trying to reach someone who is pulled so many different ways during the day (i.e. Sean) 😉.  Thinking the contact form would be the best way to start the conversation; from there, he probably will want to call you.  You're looking for the GS WiFi board + control board upgrade kit; things you would need for your unit (off the top of my head!)...

• GS WiFi board (updated to latest firmware)
  • "big" hall sensor (for WiFi board total output power monitoring)--your inverter does not have this
  • mounting screws for WiFi board (not sure if the "plate" screws on your unit are the same size?  They're M3.)
  • FDBK cable (4-pin XHB straight-through pinout i.e. 1-1, 2-2, 3-3, 4-4, NOT crossed)--your inverter does not have this
  • COMM cable (custom made for GS)
  • Battery wires for WiFi board (it directly powers from the battery terminals, due to the potential high current draw of the fans)
  • XHB-2 connectors for 2-wire fans and/or XHB-4 connectors for PWM 4-wire fans (NOTE: observe pinout!)
• GS LF Control board Rev. A.1 (latest firmware, with thermistors added, output filter caps corrected, all M5 terminal screws installed, tested)
  • NEG spade terminal and wire (long enough to reach the back negative battery terminal)
  • empty XHB-4 connector header (for the thermistors)
  • the RGB power switch + connector (this actually sorta fits into the original power switch hole in the chassis)

You'll obviously need a good bit of electrical expertise and such to get it installed.  Such as: popping the thermistor pins out of the PJ wiring harness, and slipping them into the XHB-4 connector for the GS board.  We may want to schedule a video call so I can try to make sure everything's wired up right before the dreaded "smoke test."